Blood Test Uses AI, Biomarkers to Identify Infant Heart Defects Before Birth

A study published in the American Journal of Obstetrics and Gynecology shows that a blood test, which leverages artificial intelligence (AI) and genetic biomarkers, can prenatally identify fetal congenital heart defects.

According to the Centers for Disease Control and Prevention (CDC), one in 33 infants in the US is affected by birth defects, a leading cause of infant mortality. Over 4,000 infants die annually because of congenital disabilities, and infants who survive and live with these defects are at higher risk for developing lifelong cognitive, physical, and social issues.

Congenital heart defects are the most common type of congenital disability. Statistics show that they affect roughly 1 percent, or 40,000, of births per year in the US. The prevalence of some congenital heart defects, particularly mild types, is increasing, but other types have remained stable.

As a leading cause of congenital disability-associated infant illness and death, congenital heart defects are one subset of many conditions researchers want to tackle through early detection.

“We know that when congenital heart defects are diagnosed early – ideally before birth – outcomes can improve significantly and mortality and morbidity reduced,” said Ray Bahado-Singh, MD, system chief of obstetrics and gynecology, Beaumont Health, and lead author of the study, in the press release, which was shared with HealthITAnalytics.

In this study, the researchers aimed to facilitate early detection using a minimally invasive method. They hypothesized that since epigenomics and epigenetics are crucial mechanisms for controlling gene expression in cardiac development, whole-genome and AI-based analysis had significant potential to accurately detect signs of heart defects in fetal DNA found circulating in maternal blood samples.

To test this hypothesis, the research team assessed the ability of multiple AI approaches, combined with genomic data, to identify altered gene pathways that are important in the development of congenital heart defects. They selected data from 12 cases of isolated non-syndromic congenital heart defects and 26 matched controls to evaluate the models.

The highest-performing model achieved high accuracy in detecting epigenetic changes and altered gene pathways related to the developmental processes of cardiovascular systems and function, cardiac hypertrophy, congenital heart anomaly, and cardiovascular disease. The model, which used a combination of five whole-genome biomarkers, achieved an area under the receiver operating characteristic curve of 0.97 with 98 percent sensitivity and 94 percent specificity.

These findings indicate that a minimally invasive congenital heart defect identification method could be an important first step in establishing effective postnatal action plans for at-risk infants, but larger, prospective studies are needed to validate the study’s findings, the researchers stated.

This research adds to ongoing efforts to use AI and big data analytics to tackle congenital disabilities.

In 2018, the National Institutes of Health (NIH) awarded a $10 million grant to support HudsonAlpha Institute of Biotechnology’s newborn whole genome sequencing project, designed to advance diagnosis and care of newborns with congenital disabilities and genetic disorders.

The year before, the Center for Data Driven Discovery in Biomedicine at Children's Hospital of Philadelphia (CHOP) received a $14.8 million NIH grant to explore the root causes of congenital disabilities and pediatric cancer.

Other areas of health IT, including mHealth, are being applied in similar efforts.

In 2019, Children’s of Alabama partnered with Locus Health to launch a remote patient monitoring program to monitor infants with complex heart defects following discharge from the hospital.